Detection of Early-Stage Ovarian Cancer using DNA-Decorated Carbon Nanotube Vapor Sensors

We propose an “electronic nose” technology based on arrays of carbon nanotube field-effect transistors (CNT-FETs) as a minimally invasive diagnostic for ovarian cancer. We measured the real-time electrical response of the sensor arrays to volatile organic compounds (VOCs) in the headspace of blood plasma samples collected from 21 patients with malignant ovarian cancer (six of whom had early-stage cancer), 16 patients with benign ovarian lesions, and 21 age-matched healthy subjects. Sensor arrays comprised 100 CNT-FETs, where each device was functionalized with one of ten different sequences of single-stranded DNA, providing a ten-dimensional output for each sample tested. A linear discriminant analysis tool was used to dimensionally reduce the ten-dimensional data space. A machine-learning package was then used to classify samples as malignant, benign, or healthy based on a training subset of the data, leading to correct classification of 88% of the samples in the test set. Moreover, all early-stage samples were correctly classified as malignant. This signifies a promising step towards a reliable diagnostic for ovarian cancer in the early stage, when standard therapeutic procedures are highly effective.